A scan is just a photograph of a page — Ctrl+F finds nothing. OCR adds an invisible text layer so you can search, copy, and index the document. Here's how it works and how to run it.
You scan a contract, open the PDF, press Ctrl+F to find a clause — and the search comes back empty, even though the words are right there on screen. Nothing is broken. Your scanner produced photographs of pages, and a photograph contains no text a computer can read. Making a scanned PDF searchable is exactly what OCR is for.
There are two kinds of PDF. A digital PDF — exported from Word, a browser, an invoicing tool — contains real text objects: characters with positions and fonts. A scanned PDF contains one big image per page. The quick test: try to select a sentence with your cursor. If you can highlight individual words, you have real text. If the selection is a clumsy rectangle over the whole page, you have an image-only scan — and search, copy, and screen readers all have nothing to work with.
OCR — optical character recognition — analyses the page image, recognises the shapes of letters and digits, and works out what characters they are and where they sit. The useful part is what happens next: the recognised text is written into the PDF as an invisible text layer, positioned precisely behind the printed words. The page looks pixel-for-pixel identical to your original scan, but now search finds and highlights words in the right place, copy-paste produces real text, screen readers can speak the document, and document systems can index it.
These solve different problems. OCR keeps the scan's exact appearance and adds searchable text underneath — the right choice for contracts, records, and anything where the page must remain visually authentic. Conversion rebuilds the document as editable text and layout. If you need both, OCR first, then convert the searchable result.
OCR engines use per-language models — trained character shapes plus dictionaries that help resolve ambiguous glyphs. Run an English model over French or German text and the accented characters come out mangled; run it over a different script entirely and the output is noise. Before you start, select the language your document is actually written in. For mixed-language documents, choose the language that dominates the text you care about searching. Arawa PDF's OCR tool supports the languages listed in its language picker.
Files processed on our servers are automatically deleted within one hour of processing — the searchable output is yours to download, and nothing stays behind.
Recognition quality follows scan quality. A few things make a real difference:
Once the text layer exists, the rest of the toolbox opens up: convert the document to Word at /pdf-to-word for editing, or pull the raw text out with PDF to Text for notes and quoting. Search across a folder of OCR'd scans suddenly works the way it should, and anything you file into a document system becomes findable instead of being a dead image.
The best time to OCR a scan is right after you make it, while you still know what the document is. A contract you OCR today is findable in thirty seconds next year; a folder of image-only scans is an afternoon of opening files one by one. If you're digitising paperwork regularly — receipts, records, old files — run each batch through OCR as the final step of scanning, and future-you inherits an archive that can actually be searched.